MULTIPLE FACTORS 257 
factor, there was no known process whereby a 
result of this kind would be expected. Castle found, 
on performing the experiment, that when the minus 
hooded of first extraction (—0.38) were crossed to 
normal, the minus hooded of the second extraction 
averaged +1.01, and when these were again crossed, 
the minus of third extraction averaged +2.55, and 
did not range below +1.00. The plus race, after 
the second extraction, was apparently no further 
modified than before, being +3.34, there was even 
some change in the reverse direction; but this 
result seemed due principally to the peculiar fac- 
torial composition which one aberrant family hap- 
pened to have received; after the third extraction 
further change was again noticeable, the average 
being +3.04. Moreover, some of the families were 
of almost exactly the same grade as some of those 
in the third extracted minus series. As a result of 
these experiments, Castle has reversed his earlier 
conclusion and states that the case is after all one 
of multiple modifying factors. 
In favor of the view that factors are constant are 
the convincing experiments of Johannsen on the size 
of the Princess beans. The material is highly favor- 
able for work of this kind, not only because exact 
measurements may be taken, but because the stocks 
reproduce by self-fertilization and were found to be 
homozygous. Johannsen’s results (Fig. 62) show 
that no matter how many factors influence the size 
of the bean, so long as the bean is homozygous, 
selection of plus and minus variants produces no 
